Hii Zach, I’m currently designing a circuit where I have to supply a opamp connected to 70V. The 70V is made by boosting 27V->70V. One thing is that the 70V needs to very stable(max 10mV of noise). What is the best way to get a high stable voltage? Hope to hear from you, because your explanations are always clear!
Nice video, really inspired me, I'm an electrical engineer and work on audio circuits. If I use an LC lowpass filter before my LDO, will the LDO get back the missing high frequency bandwidth lost with the LC filter ? also will the LDO act as a buffer so all the capacitors after the LDO won't interact with the LC filter before the LDO, right ?
The LDO basically acts like a resistive buffer because the function of the error amplifier in the LDO is to modulate the value of R(DS) by adjusting the gate drive voltage (assuming NMOS). So I would say yes it does block the interaction between output caps and input filter, and I think it does so by function of damping (I have never studied this function fro a circuit perspective to be honest). About the bandwidth, the bandwidth depends on the response time of the control loop to transients, I would not say that the LDO "gets back" some lost bandwidth but rather that if the input filter has too much bandwidth reduction then the control loop can become unstable, so clearly there is some limit to the bandwidth reduction or rolloff strength you can apply before the LDO control loop is unable to respond to transients anymore.
@@Zachariah-Peterson You wrote "About the bandwidth, the bandwidth depends on the response time of the control loop to transients, I would not say that the LDO "gets back" some lost bandwidth but rather that if the input filter has too much bandwidth reduction then the control loop can become unstable" But I don't get why since the LC filter is a previous transfer function before the LDO, and not being part of the feedback loop, its only interaction with it is by applying a capacitive load (and the inductance in serie with the smps I believe acts as if connected to the ground in terms of AC transfer characteristics and thus believe would be in parallel to the capacitance of the LC filter, so I guess it could indeed make a nasty resonnant filter that the Q would need to seriously study so it wouldn't go above 0dB) is this the reason why you say it could make the phase margin instable ? if this resonnance is in the open loop gain above 0dB frequencies bandwidth and not above this open loop gain I guess it would be safe, even if phase above 180° or under -180° at this particular peak frequency... ? are those condition good premises for simulations, and understanding it ?
Yeah I think that a lot of designers will just follow the application note and then say "I need a quiet supply", so they throw a filter on it not thinking about the potential for making overshoot much worse.
when I read datasheet of some pulse switch IC, sometime will see they recommend the range of ESR in output capacitor. Does it relate to keep the switching noise in constraints ?
This is done in an effort to ensure specified performance of the IC, which may include an output ripple specification. This also affects the turn-on time and the presence of any transient during turn on. In general, with higher ESR, you can get smaller transient or significantly damped transient turn-on on the output from the regulator. That is very important because some regulators can exhibit large overshoot during their turn-on phase, and this can reach significant voltage overshoot that might damage your components. So if you select the proper ESR value you can suppress this overshoot and hit a desired turn-on time.
Hi Zach, Wonderful Explanation... Just one question though.... If I make R1 0E & keep R2 as is, will that be an improvement for switching noise(ripple)???
I have not tested that so I do not know if it will create better results, I normally test these with the same values. Note that you can't ever truly get to zero because the capacitor will always have some ESR, it could be as low as 10 mOhm and as high as 500 mOhm, depending on the case size. Not all manufacturers will provide a guaranteed value for the ESR value.
Hey Zach! great job! thanks for the valuable content. testing the parameters on a real board would be really good. would love to see the results by changing some R or L or C values.
A load of 1Meg is very small so i can see the op amp method working well but at that small amount wouldn't a small LDO be better overall ? Also what is the best way to determine the desired ripple on a power line? Why was a PI filter not considered?
You can use a pi filter, same optimization method with varying parameters applies, also the final LC element later in the video acts like a pi filter directly coming off the inductor. I also wouldn't consider 1Meg to be a small load even though it should work fine for an op-amp, if you go a couple orders of magnitude smaller you will find that the converter eventually operates at or near the discontinuous mode.
Niceo video as always Sir!
Glad you liked it!
You guys should do a video on USB C
Hii Zach,
I’m currently designing a circuit where I have to supply a opamp connected to 70V. The 70V is made by boosting 27V->70V. One thing is that the 70V needs to very stable(max 10mV of noise). What is the best way to get a high stable voltage?
Hope to hear from you, because your explanations are always clear!
Nice video, really inspired me, I'm an electrical engineer and work on audio circuits. If I use an LC lowpass filter before my LDO, will the LDO get back the missing high frequency bandwidth lost with the LC filter ? also will the LDO act as a buffer so all the capacitors after the LDO won't interact with the LC filter before the LDO, right ?
The LDO basically acts like a resistive buffer because the function of the error amplifier in the LDO is to modulate the value of R(DS) by adjusting the gate drive voltage (assuming NMOS). So I would say yes it does block the interaction between output caps and input filter, and I think it does so by function of damping (I have never studied this function fro a circuit perspective to be honest). About the bandwidth, the bandwidth depends on the response time of the control loop to transients, I would not say that the LDO "gets back" some lost bandwidth but rather that if the input filter has too much bandwidth reduction then the control loop can become unstable, so clearly there is some limit to the bandwidth reduction or rolloff strength you can apply before the LDO control loop is unable to respond to transients anymore.
@@Zachariah-Peterson You wrote "About the bandwidth, the bandwidth depends on the response time of the control loop to transients, I would not say that the LDO "gets back" some lost bandwidth but rather that if the input filter has too much bandwidth reduction then the control loop can become unstable" But I don't get why since the LC filter is a previous transfer function before the LDO, and not being part of the feedback loop, its only interaction with it is by applying a capacitive load (and the inductance in serie with the smps I believe acts as if connected to the ground in terms of AC transfer characteristics and thus believe would be in parallel to the capacitance of the LC filter, so I guess it could indeed make a nasty resonnant filter that the Q would need to seriously study so it wouldn't go above 0dB) is this the reason why you say it could make the phase margin instable ? if this resonnance is in the open loop gain above 0dB frequencies bandwidth and not above this open loop gain I guess it would be safe, even if phase above 180° or under -180° at this particular peak frequency... ? are those condition good premises for simulations, and understanding it ?
Great topic for a video. One that doesn't get addressed nearly enough imo
Transient response can be a huge issue in circuits with microprocessors
Well said!
Yeah I think that a lot of designers will just follow the application note and then say "I need a quiet supply", so they throw a filter on it not thinking about the potential for making overshoot much worse.
when I read datasheet of some pulse switch IC, sometime will see they recommend the range of ESR in output capacitor. Does it relate to keep the switching noise in constraints ?
This is done in an effort to ensure specified performance of the IC, which may include an output ripple specification. This also affects the turn-on time and the presence of any transient during turn on. In general, with higher ESR, you can get smaller transient or significantly damped transient turn-on on the output from the regulator. That is very important because some regulators can exhibit large overshoot during their turn-on phase, and this can reach significant voltage overshoot that might damage your components. So if you select the proper ESR value you can suppress this overshoot and hit a desired turn-on time.
Hi Zach, Wonderful Explanation... Just one question though.... If I make R1 0E & keep R2 as is, will that be an improvement for switching noise(ripple)???
I have not tested that so I do not know if it will create better results, I normally test these with the same values. Note that you can't ever truly get to zero because the capacitor will always have some ESR, it could be as low as 10 mOhm and as high as 500 mOhm, depending on the case size. Not all manufacturers will provide a guaranteed value for the ESR value.
Hey Zach! great job! thanks for the valuable content.
testing the parameters on a real board would be really good. would love to see the results by changing some R or L or C values.
That's a really good point, I'll have to make a test board with the 12V to 5V regulator project
A load of 1Meg is very small so i can see the op amp method working well but at that small amount wouldn't a small LDO be better overall ?
Also what is the best way to determine the desired ripple on a power line?
Why was a PI filter not considered?
You can use a pi filter, same optimization method with varying parameters applies, also the final LC element later in the video acts like a pi filter directly coming off the inductor. I also wouldn't consider 1Meg to be a small load even though it should work fine for an op-amp, if you go a couple orders of magnitude smaller you will find that the converter eventually operates at or near the discontinuous mode.